 |
Liquid Cooling Plate supporting 24 7 continuous operation by effectively managing heat dissipation in gpu servers
Product Details:
| Place of Origin: | Dongguan，Guangdong,China |
| Brand Name: | UCHI |
| Certification: | UL.VDE,SGS,REACH,CQC,CSA.ISO.ROHS,CUL |
Payment & Shipping Terms:
| Minimum Order Quantity: | 1000PCS |
|---|---|
| Price: | Negotiable |
| Packaging Details: | Bulk |
| Delivery Time: | 5-7 days |
| Payment Terms: | T/T,Paypal,Western Union, |
| Supply Ability: | 5000,000,000PCS Per Month |
|
Detail Information |
|||
| Tolerance: | ±1% | Grade: | Copper |
|---|---|---|---|
| Service: | OEM Service | Precision: | ±0.002 ~ ±0.005mm |
| Heat Source Power: | 24kW | Weight: | Varies By Size, Typically 200-500 Grams |
| Product Size: | 244.5*118*14.5mm | Altitude: | ≤2000m |
| Working Pressure: | At Least 1 Bar | Base Material: | Aluminum Or Copper |
| Protection Class: | IP54 | Diameter: | Φ 8mm |
| Noise: | 17dbA | Product Weight: | 1.4kg |
| Dimension: | 268x158x22mm | ||
| Highlight: | Liquid Cooling Plate for GPU servers,24/7 continuous liquid cooling plate,GPU server heat dissipation plate |
||
Product Description
H100 / H200 GPU Liquid Cooling Assembly
Product Overview
This integrated liquid cooling module is designed for NVIDIA H100 / H200 SXM servers. The assembly includes 8 GPU cold plates and 4 NVSwitch cold plates, equipped with flexible high-pressure water hoses and color-coded quick-disconnect fittings. It is a direct liquid cooling solution for high-density AI computing servers and intelligent computing clusters. Replacing traditional air cooling, it effectively eliminates overheating and throttling issues under full heavy load, supporting stable 24/7 continuous operation.
Key Advantages
✅ Balanced Cooling Across All Chips
The cold plate channels are optimized via thermal simulation for the GPU die, HBM memory, power supply area and NVSwitch chips. The micro-channel design achieves low thermal resistance and eliminates hot spots, preventing GPU throttling to fully release computing power for AI training and large model inference.
✅ Modular Design with Low Pressure Drop
Adopts integrated manifold distribution structure with low overall pressure loss, compatible with CDU cooling distribution units in data centers. Color-coded quick connectors (red for inlet, blue for outlet) provide foolproof installation and easy maintenance.
✅ High-Reliability Sealing Process
Cold plates are made of nickel-plated oxygen-free copper for corrosion and oxidation resistance. Vacuum brazed as one piece. Every unit passes pressure holding and leak testing before delivery to eliminate leakage risks and meet strict data center operation standards.
✅ Perfect Fit for High-Density Servers
The structure matches the original mounting positions of the server. Matching pressure bars and spring fasteners are supplied together. No modification to the original server structure is required, suitable for mass retrofitting and new server deployment in standard racks.
✅ Long-Term Stable Operation
Compatible with deionized water, PG25 / PG30 coolant. The hoses feature excellent aging resistance and wide temperature tolerance. Ideal for year-round continuous workloads in intelligent computing centers, helping reduce PUE and extend GPU service life.
Specifications & Construction
- Compatible GPU: NVIDIA H100 / H200 SXM + NVSwitch
- Configuration: 8 pcs GPU cold plates + 4 pcs NVSwitch cold plates, pre-assembled
- Cold Plate Material: Nickel-plated oxygen-free copper
- Hose: Industrial high-pressure, temperature-resistant water hose
- Fittings: Industrial quick-disconnect couplings, color-coded for inlet/outlet
- Rated Total Heat Load: 6344W
- Rated Flow Rate: 9 LPM, Pressure Drop ≤ 40 kPa
- Leakage Test: 1.5 bar pressure holding test, zero leakage
- Customization Available: Channel structure, fitting type, hose length, inlet & outlet direction can be adjusted on request
Application Scenarios
🔹 Large AI model training clusters & high-performance inference servers
🔹 Intelligent computing centers, IDC data center liquid cooling renovation projects
🔹 HPC high-performance computing, scientific simulation and meteorological computing platforms
🔹 Supercomputing laboratories and private enterprise AI computing clusters
Want to Know more details about this product
